I'm a bit on the fence about it. As it stands, any HTTP library won't "just work" due to having to handle HTTPS and/or HTTP/2.0. That's not going to make people want HTTP support any less.

Then again, if you do want stdx to be something you can just blindly drop into a project... well, if they want hyper, they can always add it manually...

...except that doesn't solve the problem of people even knowing about hyper and that they should use it.

My gut tells me to keep it, and crates like it, if only to document that they are crates that you should consider using. Heck, maybe stdx could go a bit further and include a hint on what it takes to run them. Something like this added to the description:

Optional. hyper is optional because it is large, and requires: a C compiler, and the libopenssl library and headers.

That could give users a quick, rough indication as to how hard it will be to get the crate working (maybe they already have those things!).

I agree with @DanielKeep, though I am of course biased.

It's very positive to give one HTTP library a quasi-official status, keeping hyper optional in stdx is an important blessing.

OK, it seems clear that recommending a crate for important functionality is important, even if it doesn't yet warrant turning on by default.